🇺🇸 BeTraC Challenge
Location: United States
Deadline: Jul 8, 2026
BeTraC Challenge at IEEE SLT 2026 for audio AI and speech models
The Beyond Transcription Challenge (BeTraC) is a shared task at IEEE SLT 2026 that aims to address a foundational question in audio AI: Can a model reason over speech without first converting it to text?
The challenge focuses on clinical note generation from doctor–patient conversations, which is an ideal stress test for models to extract meaning directly from audio.
There are two tracks: Lightweight (≤ 6B parameters) and Heavyweight (≤ 36B parameters), with different constraints on model architecture and tool use.
The Synth-DoPaCo dataset is available on Hugging Face, and consists of 8,800 synthetic doctor–patient conversations with various ambient sounds and room reverberation.
Key dates include:
- Apr 2, 2026: Training + dev data release
- May 4, 2026: Open-source inclusion proposals deadline
- Jun 24, 2026: System description submission deadline
- ~Jul 1, 2026: Test SOAP notes due
- Jul 8, 2026: Challenge paper submission
Baselines are posted, and team registration is open. For more information, visit the BeTraC website or contact the organizers at betrac@googlegroups.com.
Tags: BeTraC, Beyond Transcription Challenge, IEEE SLT 2026, audio AI, speech models, clinical note generation, United States